BERT for Humanists

FAIN: HAA-290374-23

Cornell University (Ithaca, NY 14850-2820)
Matthew Wilkens (Project Director: June 2022 to January 2023)
Matthew Wilkens (Project Director: January 2023 to present)
Melanie Walsh (Co Project Director: June 2022 to present)
David Mimno (Co Project Director: December 2022 to present)

The development of case studies about and professional development workshops on the use of BERT (Bidirectional Encoder Representations from Transformers) for humanities scholars and students interested in large-scale text analysis.

Large language models (LLMs) have revolutionized natural language processing since their introduction in 2018. By combining huge numbers of parameters with vast text collections, pre-trained LLMs offer advanced general-purpose language understanding off-the-shelf. As powerful as LLMs can be, however, we have seen clear examples of the problems that arise from a lack of good humanities-focused resources to interpret their outputs and to guide scholars in the field. The BERT for Humanists project produces research, training, and tools to inform, empower, and inspire humanities scholars to use LLMs in their disciplines in creative new ways. Together, the products of BERT for Humanists provide an intellectual framework for understanding and evaluating new computational language technologies, so that humanists can use — and critique, as appropriate — both the current generation of LLMs and their rapidly evolving successors.